Speex dev - Jun 2007 - VAD Algorithm

Steve Underwood a ?crit :
> Coffey, Michael wrote:
>> Has it been a whole week since a question about VAD? :-)
>>
>> I am curious why the VAD algorithm described in G.729 Annex B is not
>> used by Speex. Complexity, Patent, Quality, or what?
>>   
> Its not only patented, its so darned good they recently threw it out and
> replaced it. :-) I assume the replacement is just as patented as the
> original.
In case there are volunteers, I'm interested in improvements to the
current VAD which, I have to admit, sucks quite a bit.

	Jean-Marc

I wouldn't mind taking a look at that, if you have an outline of how
you'd want it to work.

Coffey, Michael a ?crit :
> I wouldn't mind taking a look at that, if you have an outline of how
> you'd want it to work.
The only thing I have in mind is that the VAD should return 1 on speech
and 0 on noise/silence. Of course, if you insist on doing it the other
way, I'm sure we can work it out ;-)

	Jean-Marc